| 摘要: | 金融危機多年來引發了人們的關注，其對公司的影響同時震盪了全球經濟。在公司的財務決策中，其中一項非常重要的研究議題是如何準確辨識導致公司失敗的特徵。本研究使用了來自55家建築公司十六年的3190份有效財務報告，同時應用了25種財務比率，而這些比率都有著至關重要的作用。同時，本研究也提出了一種PSO-SIP演算法，將高維度的數據可視化為二維分布圖，在投影出來的分布圖中可以直接分析每個資料點群集中的固有結構，這個演算法亦可進一步結合傳統分析財務比率的方法，為決策者提供可視化圖像來為未來問題做出正確決策。此外，可視化的群集也可以協助決策者更容易理解財務比率之間的關係，並加強對它們之間相關性的研究。本研究的目的是從蒐集的資料中找出失敗的公司類別中導致公司面臨財務困境的因素，同時找出導致非失敗的公司類別成長的原因。為了實現這項研究目的，該演算法與PCA結合以確定每個特徵的權重，並在財務比率內調整和查找它們之間的關聯法則，這個方法在辨識破產分析的主要特徵時，提供了更好的可靠性。
根據本研究所使用的25個財務比率，PSOISST模型分析結果的平均準確率為90％。運用權重、調整關聯法則，該模型將資產報酬率、收入成長率、每股盈餘、淨利率、營業利潤、固定資產周轉率和借貸依賴程度作為未失敗的工程公司成長的最重要的因素。另一方面，對於失敗的公司，模型輸出八個比率：每股盈餘、資產報酬率、稅後報酬率、存貨周轉率、債務與資產比率、借貸依賴性、淨利率、營業利潤，其中的兩個比率，依賴借貸和債務與資產比率已被確定為導致公司失敗非常關鍵的因素。
企業財務困境是全球商業部門的主要關注點，因此，本研究的成果將人工智慧與統計技術相結合，希望可以提供一個改善緩解破產的方法。;Financial crisis has raised concerns for years and its effect on companies influence economies globally. The ability to accurately identify the features responsible for business failure is an important issue in financial decision-making. The study made use of 3190 effective financial reports from 55 construction companies over a decade while applying 25 ratios. All the ratios involved each play a crucial role. We proposed a PSO(SIP) algorithm to visualize high-dimensional data as a two dimensional scatter plot. The projected scatter plot allows a straightforward analysis of the inherent structure of clusters within the analyzed data points. It will also assist traditional methods in analyzing ratios by providing visualized images for decision makers to make correct decisions for future problems. In addition, the visualized clusters will provide a better understanding of the relationships among ratios and enhance the study of the correlation between them. Our goal is to determine the factors responsible for distress in the Failed category and factors responsible for growth in the non-Failed category. To achieve our goal, the algorithm is combined with PCA to determine the weights of the features and then adjust and find association rules within the ratios. This method provides better reliability in the identification of the principal features in bankruptcy analysis.
Based on the 25 ratios used, the PSOISST model yields an average accuracy rate of 90%. Applying weights, adjusting and then mining association rules, the model identified return-on-assets, revenue growth rate, earning-per-share, profit margin, operating profit, fixed assets turnover ratio and dependence-on-borrowing as the most important contributors to growth in the non-failed construction companies. On the other hand, for the companies that have failed, the model output eight ratios namely; earnings per share, return on assets, after-tax rate of return, inventory turnover ratio, debt to assets ratio, dependence on borrowing, profit margin, operating profit. Two ratios, dependence-on-borrowing and debt-to-assets-ratio have been identified as very crucial contributors to failure.
Corporate financial distress is a major concern to business sectors worldwide; therefore, combining AI with statistical techniques improves results in mitigating bankruptcy. |
